TLDR: As AI agents become integral to business operations, companies are increasingly turning to confidential computing to safeguard sensitive data. This shift is driven by the need to protect data moved by AI agents and address new security vulnerabilities, with industry leaders emphasizing the importance of robust data protection strategies and regulatory compliance.
The proliferation of AI agents across enterprise environments is ushering in a new era of data security challenges, prompting companies to rapidly adopt confidential computing solutions. This strategic pivot is deemed essential for the successful deployment and operation of agentic AI, which is poised to revolutionize internal data movement within IT infrastructures.
Industry analysts and IT executives are sounding the alarm, urging organizations to prioritize data security as AI agents take on more critical roles. The consensus is clear: neglecting data protection in this evolving landscape is no longer an option. In response, major technology players are actively developing and offering their own confidential computing frameworks to meet this burgeoning demand.
Richard Searle, Chief AI Officer at Fortanix, underscored the regulatory imperative, stating, ‘Regulation of AI use cases — such as in healthcare and financial services, [which] are already subject to extensive regulatory oversight and compliance monitoring — will require auditability of the specific AI models or agentic software deployed.’ This highlights the critical need for transparency and accountability in AI deployments, particularly in highly regulated sectors.
While AI agents promise enhanced cyber resilience and productivity, they also introduce new risks. Arvind Nithrakashyap, co-founder and CTO of Rubrik, acknowledged this dual nature, noting that ‘As AI agents become more sophisticated and interconnected, they will likely lead to more security vulnerabilities and accidental data leaks.’ He advises business and IT leaders to proactively establish stringent guardrails and data access policies to mitigate these risks.
Concerns about ‘Shadow AI’ — unauthorized or unmanaged AI deployments — are also growing. Akiba Saeedi, IBM’s VP of Security Product Management, warned that ‘Shadow AI will prove to be more common — and more — risky than we thought,’ emphasizing the need for comprehensive oversight.
The shift towards agent-driven approaches in cybersecurity is also gaining momentum. Harman Kaur, VP of AI at Tanium, predicts that by 2025, AI in cybersecurity will move beyond chatbots to more sophisticated agent-driven systems, enabling autonomous threat detection and response, and improving overall cyber hygiene. Danny Allan, CTO of Snyk, stressed the importance of embedding security from the outset, advocating for multi-model integrations that prioritize secure and efficient code generation.
Recent reports underscore the urgency of these concerns: a Stibo Systems report revealed that 75% of business leaders do not fully trust the data in their own systems due to inconsistencies and inaccuracies. Furthermore, a bigID report indicated that 70% of enterprises view data leaks by employees as a major security concern, yet only 38% of financial firms currently have an AI data protection strategy in place. The threat landscape is further complicated by cybercriminals leveraging AI models like Grok and Mixtral to generate sophisticated phishing emails and malware.
Also Read:
- Agentic AI: The Unseen Revolution Reshaping Industries
- AI Agents Reshaping Software Development and Driving Business Acceleration
The 2025 Confidential Computing Summit served as a pivotal platform for industry leaders to discuss the future of AI security and the emerging ‘Internet of Agents.’ A key takeaway from the summit was the collaborative effort among top players in AI and infrastructure to build an open, trusted Confidential AI stack with shared standards and verifiable protections. Initiatives like AGNTCY, an open-source collective by Outshift by Cisco, LangChain, and Galileo, are actively working to enable secure agent-to-agent communication at scale through pillars like discovery, composability, deployment, and evaluation, aiming to future-proof the agentic web and the trust it relies upon.
